Total Synthesis of Two Diastereomers of Megastigmane Glycoside Lauroside B

GRAPHICAL ABSTRACT Lauroside B, a megastigmane sesquiterpene glycoside isolated from L. nobilis leaves, has the potential effect of inducing apoptosis of several highly aggressive malignant melanoma cell lines. To promote the potential development of lauroside B as a possible chemotherapeutic agent to treat human melanoma, the structure and activity relationship studies should be of great importance. In this work, two diastereomers of lauroside B were synthesized through a straightforward approach and the details of the key steps were investigated, which would provide useful information for developing an efficient method toward the synthesis of lauroside B and its structural analogs.


INTRODUCTION
Natural products have proven to be a rich source of agents providing value to medicine. [1] More than half of the currently available drugs are natural compounds or are related to them structurally. [2] Natural compounds isolated from plants or other sources with diverse structures have been considered as lead compounds, and their later structural modification has afforded substances with enhanced therapeutic possibilities. [2] Sesquiterpene derivatives have been isolated from plant sources mainly, and many of them were shown to possess a range of biological and pharmacological activities. Studies into the health benefits of sesquiterpene derivatives tend to focus on their antitumor potential, as some of them have been found to show enough potential to enter clinical trials. [3,4] Less research is focused on other applications in disease treatment and on prospective health benefits. [3] Despite this, some reports indicate that there is much potential for sesquiterpene derivatives in the treatment of cardiovascular diseases [3,5] and as antimalarials and that they are responsible for a range of other effects. [3,6] In 2012, a study conducted by the Ianaro group [7] demonstrated that lauroside B (Fig. 1), a megastigmane sesquiterpene glycoside isolated from L. nobilis leaves, [8] showed a suppressing effect on the proliferation of several malignant human melanoma cell lines, including the highly aggressive A375 cell and the primary cutaneous melanoma, WM115 and SK-Mel-28. [7] This inhibition of cell

Synthesis of Diastereomers of Lauroside B 447
proliferation is due to the induction of apoptosis, which is associated with the inhibition of IκB-α degradation and constitutive NF-κB DNA-binding activity, as well as the expression, regulated by NF-κB, of two antiapoptotic genes, XIAP and c-FLIP. [7] It was highlighted that malignant melanoma is a highly aggressive tumor that frequently resists chemotherapy, [9] so the search for new agents for its treatment is of great importance. Growing evidence has implicated NF-κB [10] as an important contributor to metastasis and increased chemoresistance of melanoma. Induction of apoptosis by lauroside B in human aggressive melanoma cell lines should be considered of great biological value. To promote the potential development of lauroside B as a possible chemotherapeutic agent to treat human melanoma, the structure and activity relationship (SAR) studies should be of great importance. Herein we report our work toward the first synthetic exploration of lauroside B as well as its diastereomers.

Results and Discussion
The structure and retrosynthetic analysis of megastigmane glycoside lauroside B [1: (5S,6S,9R,7E)-megastigman-3-one-7-en-9,6-diol 9-O-β-Dglucopyranoside] [8] are shown in Figure 1. As described, 1 could be assembled by glycosylation of megastigmane acceptor 3 with O-(α-Dglucopyranosyl)trichloroacetimidate donor 2. The benzoyl (Bz) groups will be used for the hydroxyl group protection, which could be removed under mild Zemplén conditions in the late stage. [11] Furthermore, the 2-O-Bz of glycosyl donor 2 will perform neighboring group participation [12] in glycosylation, thereby controlling 1,2-trans selectivity. The glycosyl acceptor 3 could be made from nucleophilic addition of organolithium reagent 5 on the carbonyl group of intermediate 4, followed by stereoselective reduction of the triple bond on the side chain to a trans-double bond. Intermediate 4 can be delivered from commercially available 1,4-cyclohexanedione monoethylene ketal 6. This strategy seems straightforward, but how to efficiently construct the chiral centers with the desired stereochemistry is critical. The aglycone of lauroside B has three chiral centers. The R configuration at C9 can be introduced from a commercially available reagent, whereas efficient construction of contiguous chiral centers in C5 and C6 remains a challenge.
As shown in Scheme 1, the synthesis started from commercially available 1,4-cyclohexanedione monoethylene ketal 6. Direct geminal dimethylation of ketal 6 on the α-position of the carbonyl to provide 7 was achieved by treatment of 6 with sodium amide and iodomethane in toluene with a yield of 64%. Compound 7 was treated with lithium diisopropylamide (LDA) in the presence of trimethylsilyl chloride (TMSCl) to form a silyl enol ether, which then reacted with formalin and tetra-n-butylammonium fluoride (TBAF) in one pot to afford the hydroxymethylated product 4. Because compound 7 contained an axis of symmetry, hydroxymethylation on the ring destroying this symmetry gave rise to a pair of enantiomers (±)-4.
The pair of enantiomers 4 were inseparable and therefore were used as a mixture to react with lithium acetylide 5 derived from commercially available (R)-tert-butyl(pent-3-yn-2-yloxy) diphenylsilane. [13] This nucleophilic addition to carbonyl is not stereoselective and was expected to afford four diastereomers, that is, 5R/6R/9R, 5R/6S/9R, 5S/6R/9R, and 5S/6S/9R. However, when this reaction was carried out in THF at -28 • C for 3 h, only two product spots were detected by TLC, and each of them can be isolated by silica gel chromatography with the yield of 55% for 9 and 5% for 8, respectively. We envisioned that each of the above products might be a pair of two diastereomers 8a,b and 9a,b, which failed to be separated from each other by common silica gel chromatography and by recrystallization. Moreover, the two pairs of products 8a,b and 9a,b indicated the same molecular weight after analysis by ESI-MS and gave very similar NMR spectra. To prove our assumption that each product 8 or 9 contained a pair of diastereomers, we moved forward with the major product 9 and hoped that the pair of diastereomers could be separated in a late stage of the synthesis, especially after glycosylation with donor 2.
Thus, compounds 9a and 9b were subjected to reaction with chloromethyl methyl ether (MOMCl) in the presence of N,N-diisopropylethylamine (DIPEA) to afford 10a and 10b. The triple bond in 10a,b was then reduced with Red-Al [13] carefully to stereoselectively form the trans-double bond (11a,b). After removal of the silyl protecting groups on C9-OH of 11a,b, compounds 3a and 3b were provided and ready to be coupled with glycosyl donor 2. It should be noted that the 1 H NMR/ 13 C NMR spectra of each of the compounds 10, 11, and 3 showed almost only one set of signals, which could not give clear evidence that they are mixtures of diastereomers.
Interestingly and as expected, when 3a and 3b were coupled with glycosyl donor 2 promoted by trimethylsilyl trifluoromethanesulfonate (TMSOTf) in dichloromethane (DCM) at 0 • C, two glycoside isomers 12a (45% yield) and 12b (42% yield) were furnished, which were easily separated from each other by silica gel column chromatography. As mentioned above, the presence of a neighboring participation group (2-O-Bz) in the glycosyl donor 2 ensured the formation of a 1,2-trans glycosidic bond during the glycosylation. In addition, 1 H NMR spectra of the two glycosides clearly indicated that both 12a and 12b have the 1,2-trans glycosidic linkages (J H1'-H2' = 7.9 Hz and J H1'-H2' = 7.8 Hz, respectively). Thus, it could be deduced that 12a and 12b are not isomers at the anomeric center of saccharide but isomers at C5 and C6. In turn, we could confirm that compounds 9, 10, 11, and 3, as well as 8, were all mixtures of C5/C6 diastereomers.
Although the absolute configurations of C5/C6 remain unknown, we believe that the major products 9a and 9b, which were formed through nucleophilic addition of (±)-4 by lithium acetylide 5, were mixtures bearing 5S/6R and 5R/6S configurations according to the possible mechanism shown in Scheme 2. The most stable conformations of cyclohexanone are the chair forms, which are distributed in equilibrium between each other (Sch. 2). In our case, taking (S)-4 as an example, there is an additional 1,3-diaxial interaction 450 X. Fu et al.
Scheme 2: Proposed mechanism of addition of (S)-4 by alkyl lithium compound 5.
between the axial methyl group and the axial hydroxymethyl group in conformation 2 as compared with conformation 1. Thus, conformation 1 is more favored. Upon adding the alkyl lithium, the hydroxyl, carbonyl, and lithium might form a complex, a cyclic transition state, [14] which directed the alkyl addition from the axial side of the ring to provide (5S,6R)-9 as the major product (path A). Addition through conformation 2 is possible, but the hydroxymethyl group also favored directing the alkyl addition from the bottom side of the ring to provide (5S,6R)-9 (path C). By sharing the same mechanism, (R)-4 upon reacting with the lithium acetylide will provide (5R,6S)-9 as the major products. Therefore, the minor pair of products 8a and 8b was probably a mixture of (5R,6R)-and (5S,6S)-isomers. These conclusions can be supported by the products' distribution ratio of 12a:12b (≈1:1, both are from the major products mixture of 9a,b) and of the minor product 8a,b:major product 9a,b ratio (≈1:11) (Fig. 2).
Removal of ethylene and MOM protecting groups in 12a and 12b by acid followed by removing the benzoyl group via transesterification in the presence
of sodium methoxide (CH 3 ONa) provided the final products 13a and 13b, respectively. As speculated, neither 13a nor 13b is the natural product that has the 5S/6S/9R configurations, after comparison with the NMR data and optical rotation. This result further supported our proposed mechanism on the addition of (±)-4 by lithium acetylide 5 (Sch. 2). Thus, the synthesis of lauroside B should employ the (5R,6R)-and (5S,6S)-mixture 8a,b as the intermediate. However, the yield of 8a,b was low in our initial trial ( Table 1, entry 1). Several trials were made then to improve the yield of 8. Variation of the reaction temperature did change much of the total yields of the products, but an increase of the portion of 8 failed (entries 1-3). For example, at a much lower reaction temperature (entry 2), the reaction was sluggish and total yield dropped dramatically, although the portion of 8 slightly increased. A higher temperature (entry 3) resulted in the low total yields of the products, and no 8 was able to be detected by TLC and isolated. Some other reports [14a] described that in the presence of several metal salts, unusual stereoselectivity was found in the alkylation of cyclohexanones with organolithium reagents. Thus, we carried out similar reactions by adding LiClO 4 , Al(O-i-Pr), CeCl 3 , or LiBr as an additive, respectively, but all failed to get an improved yield of 8 (Table 1, entries 4-7). Other attempts using the Grignard reaction, [15] which employed the corresponding alkyl magnesium bromide instead of lithium acetylide as the nucleophile to attack the carbonyl group of (±)-4, gave similar results to those of using alkyl lithium compound.

CONCLUSIONS
In summary, we have finished the synthesis of two diastereomers of natural megastigmane glycoside lauroside B through a straightforward approach. The two diastereomers are valuable for future SAR studies. Although the current effort is not suitable to the synthesis of lauroside B, it provides useful information for developing other methods to reach the goal. Further investigations by using an alternative approach are on the way and will be reported in due course.

EXPERIMENTAL General
Products were characterized by spectroscopic data ( 1 H NMR spectra, 13 C NMR spectra, HRMS spectra). The NMR spectra were recorded on a Bruker Avance DPX 400 MHz instrument. Chemical shifts are reported in parts per million (ppm). For 1 H NMR spectra (CDCl 3 ), the residual solvent peak was used as the internal reference (7.26 ppm), whereas the central solvent peak was used as the reference (77.03 ppm) for 13 C NMR spectra (CDCl 3 ). Data are reported as follows: chemical shift, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad singlet), coupling constants (Hz), and integration. All NMR signals were assigned on the basis of 1 H NMR, COSY, and HSQC experiments. Mass spectra were recorded on an ABISciex 5800 MALDI-TOF-TOF or Shimadzu LCMS-IT-TOF mass spectrometer. The matrix used was 2,5-dihydroxy-benzoic acid (DHB) and Ultamark 1621 as the standard. Thin-layer chromatography (TLC) was performed on silica gel HF with detection by charring with 5% (v/v) H 2 SO 4 in MeOH or by UV detection. Synthesis of 7,7-dimethyl-1,4-dioxaspiro[4.5]decan-8-one (7) To a solution of sodium amide (10.0 g, 256.0 mmol) in dry toluene (50 mL) was added cyclohexane-1,3-dione 6 (10.0 g, 64.0 mmol). The reaction mixture was stirred under Ar for 30 min at rt. Then iodomethane (10.0 mL, 160.7 mmol) was added dropwise to the above reaction mixture at 0 • C and the resulting solution was stirred at rt for 4 h. The reaction mixture was quenched by addition of saturated NH 4 Cl (350 mL) aqueous solution and extracted with ether (3 × 250 mL). Combined extracts were dried (MgSO 4 ), filtered, and concentrated in vacuo. Silica gel flash chromatography afforded compound 7 (7.6 g, 64% yield). 1